/Tech21d ago

RWKV-7 G1g Launches as Top Pure RNN LLM with Fast Inference

4103226215.9K

#90

Original post

Jeremy Howard#90

BlinkDL@BlinkDL_AI

RWKV-7 G1g is here: the world's best pure RNN LLM, and a competitive LLM in general. Try https://huggingface.co/spaces/BlinkDL/RWKV-Gradio-2 for bsz16 7B inference. G1h in June 🙂 p.s. const 15000+tps decoding on single 5090: https://github.com/BlinkDL/Albatross

BlinkDL@BlinkDL_AI

RWKV-7 G1f is here (13B/7B/3B/1B) and G1g in May. p.s. Gemma 4 is great at "uncheatable eval" confirming its effectiveness 🙂 pity there's no Qwen3.5 27B base

6:27 AM · May 23, 2026 · 15.9K Views

Sentiment

Sentiment building, check back later.

Cluster Engagement

Posts from X

Most Activity

win@winwin7264

@BlinkDL_AI @JagersbergKnut Have you tried rwkv on physics of language model?

21d96

Snapolino@snapolino

@BlinkDL_AI Heya Blink, did you ever try to convert RWKV to a HRM style LLM? (slow and fast path)?

21d71